Reliability Prediction analysis is one of the most common
techniques used by engineers for reliability assessment.
Reliability Prediction provides a methodology for predicting
the failure rate or MTBF (Mean Time Between Failures) of
a product or system. Typically, Reliability Predictions are
performed early in the design cycle to evaluate predicted
product reliability. The results of the analysis can then be
used to target potential areas for improvement. This can be
critical in cases where MTBF goals are contractually
mandated or required for compliance. However, even
without contractual requirements, many organizations use
Reliability Predictions to design-in reliability in order to
ensure their products meet performance objectives.
Reliability Prediction Analysis: More Than MTBF © 2021 Relyence Corporation. All Rights Reserved. 1
WHAT ARE RELIABILITY PREDICTION STANDARDS? .................................. 2
RELIABILITY PREDICTIONS: BEYOND THE STANDARDS ............................ 2
PERFORMING RELIABILITY PREDICTION ANALYSES FASTER AND
MORE EASILY ............................................................................................................ 4
#1: EMPLOY PARTS COUNT ANALYSIS ......................................................................... 4
Using Reliability Prediction Parts Count Analysis in Early Design ..................... 4
#2: USE DEFAULT DATA VALUES ................................................................................. 6
#3: TAKE ADVANTAGE OF COMPONENT LIBRARIES ..................................................... 7
Databases of Component Failure Rates .................................................................. 8
Parts Libraries for Off-the-Shelf Components ........................................................ 9
Customized Parts Libraries .................................................................................... 9
#4: UTILIZE INTELLIGENT PART MAPPING .................................................................. 9
#5: DEVELOP KNOWLEDGE BANKS ............................................................................ 10
EXTENDING YOUR RELIABILITY PREDICTION ANALYSES ...................... 12
#1: ANALYZE OVERSTRESSED CONDITIONS WITH DERATING ANALYSIS .................... 12
#2: DETERMINE ALLOCATIONS .................................................................................. 15
#3: EVALUATE MISSION SUCCESS ............................................................................. 17
#4: CREATE AND USE FORMULAS .............................................................................. 18
#5: PERFORM WHAT-IF? TRADE-OFF STUDIES ........................................................... 20
LEVERAGING RELIABILITY PREDICTION DATA .......................................... 22
#1: INTEGRATE WITH OTHER RELIABILITY TOOLS ..................................................... 22
Reliability Prediction and RBD ............................................................................ 23
Reliability Prediction and FMECA ...................................................................... 24
Weibull Analysis and Reliability Prediction ........................................................ 24
FRACAS and Reliability Prediction ..................................................................... 25
#2: CREATE DASHBOARDS FOR OVERVIEW AND INSIGHT ........................................... 26
CONCLUSION ........................................................................................................... 29
Reliability Prediction Analysis: More Than MTBF © 2021 Relyence Corporation. All Rights Reserved. 2
Reliability Prediction analyses are performed based on a standard that offers a set of
mathematical formulas to model and calculate the failure rate of a variety of
electromechanical components. The overall failure rate of a system or product is the
summation of the underlying component failure rates. Sometimes this overall failure
rate is modified based on additional considerations, such as real-life data captured
from prior products or data accumulated from lab testing.
To learn more about Reliability Prediction, review our informative blog post.
There are several available and commonly used Reliability Prediction standards.
The most widely accepted standards are:
• MIL-HDBK-217
• Telcordia SR-332
• 217Plus
• China’s GJB/z 299
ANSI/VITA 51.1 is also a standard that utilizes MIL-HDBK-217 as a basis and
updates the failure rate calculations based on more recent device trends.
For a detailed review of Reliability Prediction standards, review our in-depth guide.
The core objective of Reliability Prediction analysis is to provide an estimate of your
product failure rate. This is an important element for reliability assessment and
continuing improvement efforts. However, the investment in Reliability Prediction
analysis can go beyond failure rate and MTBF assessments. This paper will look at
various ways you can extend your Reliability Prediction analyses to gain even more
benefit from this useful technique.
Reliability Prediction Analysis: More Than MTBF © 2021 Relyence Corporation. All Rights Reserved. 3
We will delve into the various ways Reliability Prediction analyses can be
augmented for greater efficiency and effectiveness:
• We’ll start with techniques that enable you to perform your Reliability
Prediction analyses faster and more easily.
• In the second section we’ll discuss the most common ways Reliability
Prediction analyses can be extended to include other types of analyses.
• In the last section, we’ll look at ways you can leverage the benefits of
Reliability Prediction analyses into your overall reliability and quality
program.
Reliability Prediction Analysis: More Than MTBF © 2021 Relyence Corporation. All Rights Reserved. 4
The process of performing Reliability Prediction analyses can be a cumbersome task.
Essentially, to complete an analysis, you need detailed data about your equipment
and all its components. Sometimes this information is unavailable during the early
design stage. Even if the data is available, data entry can become a tedious task.
There are several ways to lessen these burdens.
In the situation where the design is not yet complete, you can still perform
Reliability Prediction analysis. Importantly, Reliability Predictions can be very
helpful at the early design phase. Reliability Prediction helps you to recognize areas
of your design that may be problematic from a reliability standpoint. For example,
you may see that a particular subassembly or component is driving up the overall
failure rate of your design. In this case, you can reexamine the design and make
changes to stay within your reliability goals.
In the preliminary design stage, it is useful to employ the Parts Count analysis
methodology. Parts Count analyses are intended for use during the early design
stage when the overall component make up of a system may generally be known, but
specifics are not yet available. Parts Count methodology is used in contrast to Part
Stress, which is used when all or most data required for analysis is known. Not all
standards have both methodologies defined. However, several commonly used
standards, such as MIL-HBDK-217, 217Plus, and China’s GJB/z, have well-defined
Parts Count techniques.
When selecting the Parts Count analysis, significantly less data is required. For
example, here is the data requested when performing Parts Count analysis for an
Integrated Circuit when using the MIL-HDBK-217 standard:
Reliability Prediction Analysis: More Than MTBF © 2021 Relyence Corporation. All Rights Reserved. 5
Here is the same component information requested when performing Part Stress
analysis:
You can utilize the Parts Count methodology to provide a faster and easier failure
rate assessment. It also does not have to be limited to early design - you can use it at
any time if you prefer. The overall failure rate values must be considered more of a
general estimate and not a precise indicator. One example of the usage of Parts
Count methodology is to quickly compare product designs. Using Parts Count offers
a quick way to assess one design over another to determine if one design offers a
better predicted failure rate.
Reliability Prediction Analysis: More Than MTBF © 2021 Relyence Corporation. All Rights Reserved. 6
If you decide not to employ Parts Count analysis, you can still use the more common
Part Stress analysis to perform early design stage analyses by taking advantage of
default data values.
In Reliability Prediction analysis software such as Relyence, you are not required to
enter all the data parameters needed for Part Stress analyses. This is very helpful in
the early design stage when not all data parameters are known. For data values left
blank, Relyence will use built-in default values.
By design, Parts Count methodology computes failure rates by utilizing average
values for data parameters. For example, when Parts Count is in use, operating
stress values, such as power, are assumed to be 50% of maximum. Relyence Part
Stress analyses will use the Parts Count average values as the defaults for data
values not entered, allowing you to easily perform MTBF calculations with any
amount of data you have.
Relyence Reliability Prediction Part Stress uses built-in default values data not yet available.
Reliability Prediction Analysis: More Than MTBF © 2021 Relyence Corporation. All Rights Reserved. 7
This offers two additional benefits when performing Reliability Prediction analyses
with software tools such as Relyence. First, you can start off with a Part Stress
analysis, enter as much data as you have, and then go back and continually update
as information becomes available. This enables you to refine your analyses over time
as your design becomes more finalized without the need to wait until you have all
data necessary data parameters. Secondly, if you are using Parts Count in
conjunction with Part Stress, you can easily move from a Parts Count analysis to
Part Stress. As your design is refined and more data parameters are known, simply
change over to Part Stress from Parts Count and enter actual data parameters. In
either case, as more and more parameters are obtained, your Reliability Prediction
analysis becomes more accurate.
In addition, tools such as Relyence Reliability Prediction software allow you to
define your own customized default values. For example, perhaps you want to
perform early analysis using higher values than average 50% settings. Using
customizable defaults, you can change the built-in defaults to values that better suit
your requirements.
Using default values with a Part Stress methodology can greatly improve your
product design efficiency. You can begin Reliability Prediction analyses early in the
design stage before all data parameters are known and get an early indication of
potential problem areas. You then can modify and adapt your design as needed to
keep reliability goals intact. This ability to design-in reliability is one of the most
significant advantages of Reliability Prediction analysis.
One of the most time intensive elements of performing a Reliability Prediction
analysis is finding and entering all the data associated with the components of your
system. As you can see from above, you may need to enter a great deal of
information about each device. In the example above for an Integrated Circuit
calculation using MIL-HDBK-217, the package type, technology, number of pins,
number of gates, years in production, quality, and temperature are all required.
Clearly, making this data entry process most efficient is of the utmost importance.
Reliability Prediction Analysis: More Than MTBF © 2021 Relyence Corporation. All Rights Reserved. 8
The component databases NPRD (Non-electronic Parts Reliability Data) and EPRD
(Electronic Parts Reliability Data) are often used in conjunction with the Reliability
Prediction standards to augment prediction analyses. The NPRD and EPRD
databases include failure data on a wide range of electrical components and
electromechanical parts and assemblies based on actual field-based usage. You can
use this failure data in your Reliability Prediction analyses by searching for a
matching component in this large database and adding it to your analysis. In this
case, instead of component parameter data being used, the actual field-based failure
rate will be retrieved and applied.
Searching the NPRD Library in Relyence Reliability Prediction.
Reliability Prediction Analysis: More Than MTBF © 2021 Relyence Corporation. All Rights Reserved. 9
Many devices used in systems today are components manufactured by well-known
companies such as Motorola, Texas Instruments (TI), and others. The data
parameters for these components are therefore known and can be easily found.
Reliability Prediction analysis software such as Relyence offers built-in component
libraries with these widely available devices and their data parameters readily
available. For example, you only need to enter the part number “74LS00” and most
of the information needed is automatically retrieved. Design specific data, such as
operating stresses, will need to be entered; however, a large majority of the data
needed is known and easily obtained. Taking advantage of these built-in component
libraries significantly speeds up the data entry process.
In addition to commonly known devices, you may have a number of components that
you reuse in your products. In these cases, by using Reliability Prediction software
like Relyence you only need to enter the data parameters once. Once entered, you
can save information to your own Parts Libraries and retrieve it for reuse just like
built-in libraries. Customized Parts Libraries are an add-on to the built-in libraries,
allowing you to speed up the data entry process.
For maximizing data entry efficiency when using Relyence Reliability Prediction,
you can take advantage of the Intelligent Part MappingTM capability. This unique-to-
Relyence feature reviews the part description text and extracts as much information
as possible.
Reliability Prediction Analysis: More Than MTBF © 2021 Relyence Corporation. All Rights Reserved. 10
As shown above, if you enter “Cap Cer 10uF 50V” in the Description field, Relyence
will determine that the component is a 10 uF ceramic capacitor with a voltage rating
of 50 volts. Intelligent Part Mapping is one of the acclaimed capabilities of
Reliability Prediction analysis developed by Relyence. It significantly improves the
data entry process, both during direct data entry and when importing data from
outside sources.
Since its introduction, the Relyence Knowledge BankTM innovation has been widely
praised for its power, flexibility, and efficiency. Over the years it has become one of
the standout features of the Relyence platform. You can take your Reliability
Prediction analyses to a whole new level with improved productivity that can be
achieved by using Knowledge Banks.
The core function of the Knowledge Bank is to enable you to capture and reuse your
Reliability Prediction data. For example, you may have a subassembly used across
multiple product lines. Or you may use a subassembly from a prior design in a new
design. In these cases, the Knowledge Bank replaces the cumbersome and error-
prone copy-and-paste method of managing reused data. Essentially, you store the
information once in the Knowledge Bank and then reference it where needed.
Beyond the aspect of reuse without data reentry, the Knowledge Bank also offers the
ability to keep modifications made to the stored data in sync wherever used. For
example, perhaps you have changed suppliers for a particular component in one of
your subassemblies. When using the Knowledge Bank, you simply change this
information in one place. The updated information can then be pushed out to all
applicable analyses. You can choose how and when these push updates occur so that
you can control and manage the data update process.
Using the Knowledge Bank in Reliability Prediction for the management and control
of reusable data is a simple three-step procedure:
1. Add the subassembly to the Knowledge Bank.
2. Reference the stored subassembly in all analyses where it is used.
3. Keep changes made to the subassembly in sync across all analyses by using
the Knowledge Bank Push feature.
Reliability Prediction Analysis: More Than MTBF © 2021 Relyence Corporation. All Rights Reserved. 11
It is easy to see how the value of the Knowledge Bank grows over time. As you
continue to add to your bank of reusable assemblies, your future analyses can be
completed much more efficiently. Not only does it improve the speed of performing
your Reliability Prediction analyses, but it also means you have more confidence in
the consistency of your data. Because the Knowledge Bank provides a central
repository for information that is controlled and managed, you do not have to worry
about keeping track of various “versions” of your data.
The Relyence Knowledge Bank has an array of capabilities including:
• You can have multiple Knowledge Banks and store and retrieve data from
any or all.
• You can control all aspects of changes made via the Push function. You can
see a list of pending push changes and opt to bypass or delay as needed.
• You can enable notifications to allow team members to be notified of Push
changes.
• You can view a list of Knowledge Bank data usage to see analyses where data
is tied to a Knowledge Bank.
• You can extend Audit Trail tracking to Knowledge Banks.
For more information on using Knowledge Banks in Relyence Reliability Prediction
with an example walkthrough, see our instructive blog post.
Reliability Prediction Analysis: More Than MTBF © 2021 Relyence Corporation. All Rights Reserved. 12
Now that we have described the methods that enable you to perform your Reliability
Prediction analyses more efficiently, let’s look at ways they can be extended for more
in-depth studies. These add-on analyses can be done based on the information
already available in Reliability Prediction.
Derating is a means to prolong a component’s life by ensuring it operates below its
recommended maximum rated capacities. On component data sheets, electronic
device manufacturers include maximum ratings for common stresses such as power,
Reliability Prediction Analysis: More Than MTBF © 2021 Relyence Corporation. All Rights Reserved. 13
current, or voltage. By operating at less than the maximum stresses, devices last
longer. This means that failures occur less frequently, and therefore, product
reliability is improved.
Part of the information required to perform Reliability Prediction analysis is the
operating stresses of the system’s components. Combining this available information
with the components’ derating profiles enables you to perform derating analysis
alongside your Reliability Prediction analysis. Derating analysis evaluates the
operating stresses components are subjected to and then identifies those that are
above acceptable levels.
There are derating standards which are commonly used to perform derating analysis
– these include:
• MIL-STD-1547
• MIL-STD-975M
• TE000-AB-GTP-010
These standards include derating profiles for a wide range of devices. A derating
profile defines the stress ranges within which a device should operate. Operating
outside the ranges defined in the derating profile results in an overstressed device.
The characteristics of a derating profile vary. For example, a derating profile may be
a simple value: a specific type of resistor operating over 75% of maximum power is
considered overstressed. Or it may be more complicated, as shown in the derating
curve profile below. Using the nominal derating curve, the device is overstressed if
operating above 80% stress when below 25 degrees. At 25 degrees, the overstress
level varies linearly depending on temperature. At 50% stress the device is not
overstressed until the temperature is over 50 degrees. It is also common to consider
both nominal and worst-case scenarios when performing derating analysis.
Reliability Prediction Analysis: More Than MTBF © 2021 Relyence Corporation. All Rights Reserved. 14
Example derating curves for nominal and worst-case overstress conditions for a capacitor.
Relyence Reliability Prediction automatically performs derating analysis during
failure rate calculations. Any components exceeding their stress limitations are
flagged and appear in red on your tables and reports. Using the Relyence Pi Factors
feature, you can view the specific overstress condition for any component in your
analysis.
In addition to supporting the commonly used derating standards, you can define
your own completely custom derating profiles in Relyence Reliability Prediction.
Reliability Prediction Analysis: More Than MTBF © 2021 Relyence Corporation. All Rights Reserved. 15
Specific overstress conditions are noted on the Pi Factors dialog in Relyence Reliability
Prediction.
Reliability allocation is a process used to optimally distribute reliability goals across
a system in order to meet overall reliability objectives. For example, system
integrators may have a specific reliability goal either internally driven or
contractually mandated. In this situation, it is important to effectively distribute
that goal down to subcontractors and suppliers in a reasonable fashion. By
establishing reliability goals for each subcontractor, or each subsystem, a system
integrator can be assured that the overall objective will be achieved.
There are several proven allocation methods used to perform reliability allocation
analysis. Which you select depends on the data you have as well as your
determination as to how best to allocate goals.
Reliability Prediction Analysis: More Than MTBF © 2021 Relyence Corporation. All Rights Reserved. 16
Allocation methods use various ways distribute the reliability goal based
on weighting factors. Weighting factors allow for various measures to influence how
much or how little of the overall goal to assign to each subsystem.
Relyence Reliability Prediction supports the following methods for allocation
analysis:
• AGREE: The Advisory Group on Reliability of Electronic Equipment
(AGREE) developed this allocation method. An Importance Factor is used as
the basis for the weight factor for Allocation analysis. The Importance Factor
accounts for the complexity of each subsystem (based on number of
components in the subsystem) and its overall importance to the system.
• ARINC: The failure rate of each subsystem is entered and used to determine
the weighting factor for Allocation analysis. Typically, the predicted failure
rates are used, but specified values can be entered. The allocation goal is
distributed across subsystems based on their associated portion of the overall
predicted failure rate.
• Equal Apportionment: The goal is distributed equally across all subsystems;
therefore, no weight factors are required.
• Feasibility of Objectives: The weight factor for allocation is determined based
on 4 rating factors: Intricacy, State of the Art, Performance, and Environment.
• Repairable Systems: The availability goal is distributed across all subsystems
using MTTR values as the weighting factors.
• Weighted: The goal is distributed equally across all subsystems based on the
weight factor entered.
Reliability Prediction Analysis: More Than MTBF © 2021 Relyence Corporation. All Rights Reserved. 17
Performing allocation calculations in Relyence Reliability Prediction.
When performing allocation analysis, the results allow you to pinpoint areas where
allocation goals are a potential issue. Based on your allocation parameters, the
calculation will determine the allocated failure rate for each subassembly in your
system. Using the predicted failure rates from your Reliability Prediction analysis,
you can view which subassemblies are likely to exceed their allocated failure rate
goals. Armed with this knowledge, you can decide on an action plan to maintain your
overall reliability objectives. Perhaps you determine a particular subassembly needs
some redesign work, or perhaps you decide to change your allocation parameters in
order to reallocate failure goals.
By default, Reliability Prediction analysis is performed on a system assuming a
defined environment and at a defined temperature. In most cases, this is sufficient,
especially in cases where a product is used in consistent environmental conditions.
However, in some cases you may want to assess performance over more than one
operating condition.
Reliability Prediction Analysis: More Than MTBF © 2021 Relyence Corporation. All Rights Reserved. 18
One common extension of Reliability Prediction MTBF analysis is to assess mission
performance. Mission profiles help to analyze in the situations where multiple
environmental conditions are experienced. The fundamental premise of mission
profile analysis is to determine the likelihood of mission success, which is critical in
many situations. For example, the mission of an aircraft would be to takeoff, reach
the target location, and land without failure.
When looking at system performance over a mission, factors that affect reliability
vary and need to be accounted for. Some examples of these factors include the
environment, temperatures, and stresses. There may even be dormant phases to
take into account.
Example mission profile for an aircraft.
In Reliability Prediction, you can define your Mission Profile, or a listing of the
various phases of the mission you want to assess. Using the MTBF equations from
the Reliability Prediction standards, the mission-based metrics are computed. The
results provided include the predicted failure rate and MTBF for each mission phase
as well as the entire mission.
As with any analysis tool, there are times when you want to extend the analysis to
include more than what is provided out-of-the-box. For example, there may be other
calculated values you want to see in your Reliability Prediction analyses. Or you
Reliability Prediction Analysis: More Than MTBF © 2021 Relyence Corporation. All Rights Reserved. 19
may have you own set of metrics your organization wants to track. The Formulas
capability in Relyence Reliability Prediction enables you to define you own set of
computations to perform using your Reliability Prediction data.
Formulas enable you to harness the power of your collected data. When performing
Reliability Prediction analyses, a great deal of data is collected and analyzed in
order to calculate predicted failure rates. Given this wealth of information, you can
create your own Formulas to compute metrics that may be more helpful for your
specific needs or allow for more insight into your design. A common use of Formulas
is to adjust predicted failure rates from the standards-based results that align with
your field-based experience.
In Relyence Reliability Prediction, you use an intuitive Formula Builder to create
your own custom calculations. You can create any number of customized
computations using a wide variety of available mathematical functions.
Additionally, roll-up and aggregate functions such as SUM (summation) and AVG
(average) are available that allow for the computation of cumulative metrics. Also,
results from one Formula can be used in another Formula to create even more
complex calculations.
In addition to arithmetic operators and mathematical functions, Formulas also offer logical
operations to build complex and useful custom computations.
Reliability Prediction Analysis: More Than MTBF © 2021 Relyence Corporation. All Rights Reserved. 20
The power and flexibility provided with Formulas significantly advances your
Reliability Prediction analyses beyond the standards-based failure rate and MTBF
metrics.
One of the key benefits of employing Reliability Prediction analysis is to evaluate
design alternatives and their effect on product reliability. Therefore, you can use
Reliability Prediction to make crucial design decisions early on to ensure the
resulting product design will meet your reliability goals. In this light, you can
understand why Reliability Predictions can play an especially useful role in early
design prior to product manufacturing.
To utilize Reliability Predictions for this purpose, What-If? analyses are uniquely
helpful. What-If? analyses are a built-in, efficient way to compute MTBF based on
different versions of a product design. For example, you may be performing a
Reliability Prediction analysis based on a product in the design stage and have
calculated the overall system predicted failure rate. At this point you would like to
see the impact of procuring certain devices at a different quality level. Perhaps you
want to lower the cost by purchasing lower quality parts and want to see how that
affects your failure rate. Or maybe you want to improve your product’s predicted
MTBF, so you want to view the impact of using higher quality components. With
What-If? analyses you can make these changes with a few simple clicks and
recalculate to view the resulting failure rate values. These results can be used to
help guide design reviews and your decision-making process.
Reliability Prediction Analysis: More Than MTBF © 2021 Relyence Corporation. All Rights Reserved. 21
What-If? studies can help you evaluate design alternatives.
Relyence Reliability Prediction includes a streamlined What-If? Scenario Editor for
quick and easy assessment of trade-off analyses. You can create any number of
trade-off studies by changing a variety of data parameters such as temperature,
environment, quality level, duty cycle, specified failure rate, and calculation model.
The changes can be applied across your entire system, to specified system
components, or even at the part level. The flexible interface keeps all your What-If?
MTBF calculation analyses organized and on-hand for further review and
refinement at any time during the design process.
In addition, the Relyence Reliability Prediction Dashboard includes a What-If?
widget. This visual view of your What-If? scenario metrics provides an impressively
powerful way to see the effects that various design decisions have on product
reliability.
Reliability Prediction Analysis: More Than MTBF © 2021 Relyence Corporation. All Rights Reserved. 22
We have looked at ways that allow you to perform your Reliability Prediction
analyses more efficiently and extend analysis beyond the assessment of failure rate
values. Another advantage of Reliability Predictions is the ability to leverage your
Reliability Prediction data in other aspects of your reliability platform. In this way,
Reliability Prediction becomes an important component of your reliability toolset.
Reliability Prediction data can be utilized in your other reliability analysis tools.
This data sharing means your overall analysis platform is more cohesive and offers
greater efficiency and productivity. Because different analysis techniques are often
performed by different teams, this integration allows your team to work together
and recognize how everyone is invested in the same overall reliability goals.
Reliability Prediction Analysis: More Than MTBF © 2021 Relyence Corporation. All Rights Reserved. 23
Reliability Prediction analyses are oftentimes coupled with Reliability Block
Diagram (RBD) analyses to allow for a broader, more complete system profile. In
Reliability Prediction analysis, it is assumed that all subassemblies are connected in
series. However, your overall system model may include branches to connect
subsystems in a variety of ways or may include redundant components.
Redundancy is employed in system design to ensure that when a component or path
fails, a secondary component to path can take over to keep the system operational.
One of the main benefits of RBD analysis is that it allows you to perform reliability
and availability calculations for systems that include redundant components.
In cases where your system design extends beyond a simple series configuration,
RBD analysis can be key for full system performance analysis. In these situations,
Reliability Prediction and RBD often go hand in hand. Using the two together
enables you to perform complete reliability and availability analyses on your system.
Use Reliability Prediction together with RBD for a powerful reliability analysis platform.
Reliability Prediction Analysis: More Than MTBF © 2021 Relyence Corporation. All Rights Reserved. 24
To use the two tools together, you can use the failure information from your
Reliability Prediction analysis when creating your system model in RBD. In this
way, your analyses are tied together. If you make design changes which impact the
failure rate data computed in Reliability Prediction, those changes will
automatically be taken into account in your RBD analysis.
The Reliability Prediction-RBD duo provides a superior approach for comprehensive
system modeling.
FMECA, or Failure Mode, Effects and Criticality Analysis is a highly detailed type
of FMEA (Failure Mode and Effects Analysis). FMECAs incorporate the calculation
of criticality values down to the piece-part level. Failure rate values are used to
perform this criticality evaluation. The most efficient way to obtain failure rate data
required in FMECAs is by using the values obtained from Reliability Prediction
analysis. For this reason, FMECAs are often coupled with Reliability Prediction.
FMECAs typically follow the methodology defined in MIL-STD-1629A. Oftentimes,
these FMECAs are done in conjunction with MIL-HDBK-217 Reliability Prediction
analyses. The origin of both these standards is the US Department of Defense, so
military applications often use these analyses together. However, due to the wide
acceptance of these standards, other industries have also adopted this approach for
comprehensive failure analysis.
When using both Relyence Reliability Prediction and Relyence FMECA, failure rate
values are automatically shared to streamline data handling and provide a seamless
integration between your analysis tools.
Reliability Prediction analyses provide predicted failure rate values. These values
are derived from the statistical analysis of a large quantity of data over a long period
of time which resulted in the equations in the various Reliability Prediction
standards. These models are therefore based on probabilities as determined from
historical data.
Reliability Prediction Analysis: More Than MTBF © 2021 Relyence Corporation. All Rights Reserved. 25
In contrast, Weibull Analysis techniques are employed to perform a similar type of
probabilistic assessment based on real world accumulated data. When using Weibull
Analysis, you enter data acquired during actual product use, such as failure times,
number of failures, and number of non-failures. Weibull Analysis incorporates a
number of statistical techniques based on distribution analysis to curve fit the data
provided in order to evaluate data trends and product performance.
You can combine these two techniques to incorporate your Weibull-based analyses in
with your Reliability Prediction analyses. In Relyence Reliability Prediction, you can
directly link data from Relyence Weibull to components in your Reliability
Prediction analyses. The incorporation of this field-based data into your Reliability
Prediction analyses offers even greater confidence in your reliability assessments.
FRACAS, or Failure Reporting, Analysis, and Corrective Action System, is a
management system for handling issues that arise with products or processes.
FRACAS falls under the broad umbrella of Corrective and Preventive Action (CAPA)
processes which provide a framework for the effective handling of any type of failure,
complaint, incident, issue, problem, or concern. The main benefit of CAPA and
Reliability Prediction Analysis: More Than MTBF © 2021 Relyence Corporation. All Rights Reserved. 26
FRACAS is the assurance that as incidents arise, they are captured and tracked
until properly addressed, or “closed out” in CAPA terminology.
Because FRACAS captures real world data such as product failures and repairs,
there is a wealth of information available. Similar to Weibull, the real-world data
from your FRACAS can be integrated into your Reliability Prediction analyses.
Using Relyence, this is a two-step process involving three products of Relyence
Reliability Studio. From Relyence FRACAS you can create a Weibull Data Set based
on accumulated failure data. As described above, this Weibull Data Set can then be
linked to your Reliability Prediction analysis. The ability to combine predictive
measures of Reliability Prediction with real-world metrics from FRACAS provides a
more finely tuned evaluation of product performance.
The results of Reliability Prediction analyses are most commonly a list of failure
related metrics, such as failure rate and MTBF. Though these resulting metrics can
be viewed on screen and in a variety of reports, this may not provide the clearest and
most concise way to interpret the data. This is where dashboards play an important
role for quality data analysis.
In general, dashboards are a way to accumulate a variety of data into a cohesive
visual presentation. The most common dashboards we encounter are in our cars.
They allow us to quickly review all the most important information at a glance. In
the same way, dashboards are used in software applications in order to present vital
information in an easy-to-read overview.
The Relyence Reliability Prediction Dashboard gathers and organizes your data to
provide a holistic overview of your product or system. As one example, you can use
the Relyence Reliability Prediction Dashboard to view all overstressed components
in your design.
While the Relyence Reliability Prediction Dashboard provides a high-level overview,
the underlying data is always available at your fingertips with the click of your
mouse. The Relyence Reliability Prediction Dashboard’s drilldown feature takes you
from a chart, table, or graph directly to the corresponding analysis information. For
example, perhaps you want to delve into the details about the component in your
system with the highest failure rate. By clicking that part on the dashboard, the
Reliability Prediction Analysis: More Than MTBF © 2021 Relyence Corporation. All Rights Reserved. 27
Reliability Prediction analysis will automatically be opened, and you will be taken to
the part in question where you can view detailed data such as operating
temperature, stress, etc.
Dashboards provide an efficient and organized high-level overview of performance metrics.
Dashboards are customizable, so you can select the elements which are most useful
for your requirements. Additionally, with Relyence Dashboards, you can mix results
across various products for even greater flexibility and insight. For example, you can
view results from Reliability Prediction analyses alongside results from RBD
analyses on the same Dashboard.
Some of the elements that are available on a Relyence Reliability Prediction
Dashboard include:
• Failure Rate Percentage - a pie chart showing the failure rate percentage of
each subsystem in your system.
• Failure Rate vs Temperature - a line chart of failure rate values over a
specified range of temperatures.
Reliability Prediction Analysis: More Than MTBF © 2021 Relyence Corporation. All Rights Reserved. 28
• Overstressed Parts - a listing of overstressed parts.
• Parts Contributing 80% of Failure Rate - bar chart showing the parts
whose failure rates contribute to 80% of the overall failure rate.
• Top (N) Failure Rate Parts - a bar chart of the parts with the (N) number
highest or lowest failure rate values.
• What-If? - a bar graph of showing a comparison of failure rates of specified
What-If? scenarios.
The Relyence Reliability Prediction Dashboard is useful for any or all team members
such as high-level managers, analysts, engineers, and designers. As with all well-
designed dashboards, the Relyence Reliability Prediction Dashboard allows team
leaders as well as contributors to gather and react to information in real-time to
keep quality and reliability objectives on track.
Reliability Prediction Analysis: More Than MTBF © 2021 Relyence Corporation. All Rights Reserved. 29
Oftentimes, engineers approach Reliability Prediction as a singular tool used to
provide failure rate assessments of electromechanical systems. And often they view
the Reliability Prediction process as time-consuming and cumbersome. Both are
misconceptions that can be easily dispelled by taking a more in-depth look as we’ve
done in this article. Your investment in Reliability Prediction analysis can be
leveraged in a variety of ways to advance and enhance your continuous improvement
efforts.
First, there are numerous ways to improve data entry efficiency while at the same
time improving analysis effectiveness, such as utilizing Parts Count when
applicable, using default values and Intelligent Part Mapping, taking advantage of
device libraries including NPRD and EPRD databases, and leveraging the power of
Knowledge Banks.
Secondly, you can extend your analyses for further gains based on your initial
investment in Reliability Prediction analysis by using advanced techniques
including derating analysis, allocations, formulas, mission profile analysis, and
What-If? studies.
Lastly, leverage Reliability Prediction as part of your complete reliability analysis
toolset with the use of Dashboards for helping design decisions and integrations
across your other reliability tools for accurate system performance assessments.
At Relyence our mission is
not only to create the best-
in-class toolset, but also to
offer a robustness and
design elegance above and
beyond expectations for
analytical software tools.
We believe we’ve achieved
our objectives with
Relyence Reliability Prediction. We hope that you take the opportunity to give us a
free trial test run. See Relyence in action for yourself or allow us to guide your
through a personalized demo on your schedule or feel free to contact us.